On Wed, Mar 9, 2016 at 4:12 AM, Fabien COELHO <coelho@cri.ensmp.fr> wrote:
>> On Wed, Jan 27, 2016 at 2:31 PM, Fabien COELHO <coelho@cri.ensmp.fr>
>> wrote:
>>>
>>> - when a duration (-T) is specified, ensure that pgbench ends at that
>>> time (i.e. do not wait for a transaction beyond the end of the run).
>>
>>
>> Every other place where doCustom() returns false is implemented as
>> return clientDone(...). I think this should probably do the same.
>
>
> Why not. clientDone() second arguments is totally ignored, I put true
> because it looks better.
>
>> I also think that we should probably store the end time someplace
>> instead of recomputing it in multiple places (this patch adds two such
>> places).
>
>
> Why not.
>
> Attached is a v4.
OK, I've committed the fix for the -T part. It didn't back-patch
cleanly, and it is a minor bug, so I'm not inclined to worry about it
further.
I didn't commit the fix for the -P part, because Alvaro objected to
the proposed method of fixing it as ugly, and I think he's right.
Unless you can come up with a nicer-looking fix, I think that part is
going to stay unfixed.
--
Robert Haas
EnterpriseDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company